The cast: named agents

Inside the Agent Store, every 10ex agent shows up as a named character with a level, a specialty, and a per-run credit cost. The character names aren’t decoration. They’re how the product surfaces an agent in the sidebar, in chat, in the Launchpad action plan, and in run history. Hiring an agent puts that character on your team.

The roster

Two characters share names (Marcus covers Blog and Prospector; Nova covers Email Sequence and PPT). The product disambiguates by slug in URLs and role label in the sidecar. Pricing and behavior are tied to the slug, not the name.

Levels

Every character carries a level that signals the kind of work they take on.

• Junior. Single-task executors. Take a brief, return one artifact. Examples: Luna (ad copy), Orion (one cold email), Iris (one webinar image), Nova/PPT (a deck), Kai (one outbound call).
• Manager. Multi-step coordinators that run a small team of sub-agents. Examples: Nova/Email Sequence (6 sub-agents), Marcus/Blog (research, outline, draft, visuals), Atlas (outline, script, slides, video, stitch), Juno (market research with deliverables).
• Associate. Strategy and planning. Output is a plan, an audit, a structure, not the final artifact. Examples: Maya (cross-platform ads strategy), Sora (growth audit), Marcus/Prospector (Apollo search planning plus enrichment).

The level shows in every Agent Store card and in the active-agents sidebar. Use it to decide whether you want a deliverable now (Junior), a managed workflow (Manager), or strategic input you’ll act on yourself (Associate).

Credit cost

Costs are per run, charged from your workspace credit balance.

• Free agents (Sora, Luna, Dante, Zara, Ravi, Elena, Juno, Atlas, Maya). Strategy, planning, and single-image work that runs on platform LLMs you’ve already paid for.
• 1 credit. A single billable unit of output (Marcus/Prospector pulls one batch of leads; Kai makes one call).
• 1 credit per email. Nova and Orion charge per email actually sent, not per sequence generated. Preview is free; approval costs nothing; the meter starts on send.
• 10 credits. Heavy generative work (Marcus/Blog full pipeline; Iris webinar hero image; Nova/PPT branded deck via Gamma).

See Credits & pricing for tier-level math.

Why personas

Naming agents isn’t whimsy. It solves three concrete problems.

1. Disambiguation in run history. “Nova ran at 9:14am” is unambiguous. “The email agent ran” is not, when you have Nova and Orion both touching email.
2. Sidebar density. The Active Agents view shows 16 cards. A name plus avatar reads in 200ms; a slug does not.
3. Conversational handoff. Unified Chat routes messages to a specific agent. “Ask Marcus to draft a blog about X” works because Marcus is a referent, not a category.

Where personas show up

• Agent Store (/workflows/agents): the hire page.
• Active Agents (/agent-hub/active-agents): your hired team, with credit cost, latency, and total runs.
• Launchpad (Dashboard, Launchpad tab): recommended agents for each step of your 30-day roadmap.
• Unified Chat (/chat): each conversation is locked to one agent at a time.
• Run logs and Langfuse traces. The persona name flows through to observability.

Common questions

Why are two Marcus agents named the same thing? They share a personality (analytical, methodical) but cover different surfaces. The slug disambiguates: blog-generation vs prospector-agent. Pick by the surface you need.

Can I rename a persona? Not in v1. Names are global to the product so docs, run history, and chat handles all line up.

Which persona should I hire first? It depends on your motion. For outbound-heavy teams, Marcus/Prospector and Nova. For content-led teams, Marcus/Blog and Atlas. Launchpad picks for you if you don’t want to choose.

Related

• Agents architecture: what runs underneath the persona
• Brand knowledge schema: what each agent reads before it works
• Launchpad: strategic playbook: how the platform sequences personas across a 30-day plan